U.S. patent application number 12/032819 was filed with the patent office on 2009-07-23 for system and method for automatically classifying search results.
This patent application is currently assigned to Sungkyungkwan University Foundation for Corporate Collaboration. Invention is credited to Hyunseung Choo, Hyungsuk Ji.
Application Number | 20090187548 12/032819 |
Document ID | / |
Family ID | 40877239 |
Filed Date | 2009-07-23 |
United States Patent
Application |
20090187548 |
Kind Code |
A1 |
Ji; Hyungsuk ; et
al. |
July 23, 2009 |
SYSTEM AND METHOD FOR AUTOMATICALLY CLASSIFYING SEARCH RESULTS
Abstract
Disclosed is a system and a method for automatically classifying
search results. The system includes a search engine server for
obtaining and providing search results with regard to a search word
entered by the user, grouping the obtained search results according
to meanings of the search word, and providing the grouped search
results; a related word database for storing related words
classified into groups according to meanings of the search word;
and a group determination system for receiving search results from
the search engine server, comparing the contents of the search
results with the related words stored in the related word database
to determine which group of the related word database the search
results belong to, and storing search results at a predetermined
place when the search results are not grouped.
Inventors: |
Ji; Hyungsuk; (Seoul,
KR) ; Choo; Hyunseung; (Seoul, KR) |
Correspondence
Address: |
RATNERPRESTIA
P.O. BOX 980
VALLEY FORGE
PA
19482
US
|
Assignee: |
Sungkyungkwan University Foundation
for Corporate Collaboration
Seoul
KR
|
Family ID: |
40877239 |
Appl. No.: |
12/032819 |
Filed: |
February 18, 2008 |
Current U.S.
Class: |
1/1 ;
707/999.004; 707/E17.017 |
Current CPC
Class: |
G06F 16/9577
20190101 |
Class at
Publication: |
707/4 ;
707/E17.017 |
International
Class: |
G06F 7/06 20060101
G06F007/06; G06F 17/30 20060101 G06F017/30 |
Foreign Application Data
Date |
Code |
Application Number |
Jan 22, 2008 |
KR |
10-2008-0006578 |
Claims
1. A system for automatically classifying search results, the
system comprising: a search engine server for obtaining and
providing search results with regard to a search word entered by a
user, receiving grouped search results, and providing the user with
the grouped search results; a related word database for storing
related words classified into at least two groups according to
meanings of the search word; and a group determination system for
receiving search results from the search engine server, comparing
the search results with the related words stored in the related
word database to determine which group of the related word database
the search results belong to, providing the search engine server
with grouped search results, and storing search results at a
predetermined place when the search results are not grouped.
2. The system as claimed in claim 1, further comprising a category
determination system for classifying the search results stored at
the predetermined place without being grouped by the group
determination system according to domain names providing the search
results.
3. The system as claimed in claim 1, wherein the search engine
server comprises a search result query processor for querying the
group determination system regarding which group the search results
corresponding to the search word entered by the user belong to.
4. The system as claimed in claim 1, wherein the group
determination system comprises: a count processor for counting how
many related words stored in a plurality of groups constituting the
related word database are contained in contents of the search
results group by group; a group allocation processor for
determining which group of the related word database the search
results belong to according to a number counted by the count
processor; and a non-group allocation processor for identifying
search results not being grouped and storing the identified search
results at a predetermined place.
5. The system as claimed in claim 1, wherein the group
determination system comprises: a count processor for counting how
many related words stored in a plurality of groups constituting the
related word database are contained in contents of the search
results group by group; an intelligent group decision processor for
assigning weights to respective related words stored in the related
word database according to a degree of correlation between the
related words and meanings of the search word; a group allocation
processor for combining the weights assigned to the related words
by the intelligent group decision processor with numbers counted by
the count processor to determine which group of the related word
database the search results belong to; and a non-group allocation
processor for identifying search results not being grouped and
storing the identified search results at a predetermined place.
6. The system as claimed in claim 2, wherein the category
determination system comprises: a domain group database for storing
domain names classified hierarchically into at least two groups;
and a category-based search result allocation processor for
classifying the search results by determining which group of the
domain group database a host belongs to according to a domain name
of the host, the host having provided the search results.
7. The system as claimed in claim 1, further comprising: a user
preference acceptance system for enabling the user to set the
number of columns, the grouped search results being outputted in
the columns, and redisplaying the search results according to the
user setting or displaying next search results according to the
user setting; and an initial column number update system for
statistically surveying user preferences regarding the number of
groups, the search results being classified into the groups, and
automatically updating an initial setting according to the user
preferences.
8. A method for automatically classifying search results with
reference to a related word database storing groups of words
related to a search word, the method comprising the steps of: (a)
receiving a search word entered by a user; (b) obtaining search
results with regard to the entered search word; (c) grouping the
search results by a group determination system with reference to
the related word database; and (d) sorting and providing the
grouped search results in at least two groups.
9. The method as claimed in claim 8, wherein, in step (c), weights
are assigned to respective related words belonging to each group of
the related word database according to a degree of correlation
between the related words and meanings of the search word, each
weight is combined with a number of appearance of the related word
in the search results, and the search results are allocated to a
group having a high combined weight.
10. The method as claimed in claim 8, further comprising a step of
(e) repeating steps (c) and (d) when the user wants to divide some
groups of the grouped, outputted search results into subgroups and
terminating searches when the user does not want to divide some
groups of the grouped, outputted search results into subgroups.
11. The method as claimed in claim 10, further comprising a step of
(f) classifying the search results according to a group of a domain
group database, a domain name providing the search results
belonging to the group, by a category-determination system when it
is considered impossible to group the search results by the group
determination system, and outputting the classified search results.
Description
FIELD OF THE INVENTION
[0001] The present invention relates to a system and a method for
automatically classifying search results, and more particularly to
a system and a method for automatically classifying search results,
wherein, in the case of a search word which is ambiguous or which
has different meanings depending on the context, reference is made
to a related word database storing groups of words related to
respective meanings of the search word, the related words are
compared with the contents of the search results, the search
results are grouped, and the grouped search results are sorted in
at least two columns and then outputted.
BACKGROUND
[0002] As generally known in the art, recent development of
information technology and widespread use of the Internet have
enabled users to easily access a large amount of information.
However, when a user has accessed such a large amount of available
information, it also includes some pieces of information the user
does not want. For these reasons, users have tried to find a way to
search desired information only in a fast and efficient manner, and
search engines have appeared to satisfy such demands.
[0003] In general, search engines refer to search systems employed
by users to find information necessary to solve given problems.
Search engines are used to conduct searches through the Internet or
networks, desktop searches in PC or laptop environments or in other
storage spaces, and searches based on mobile devices (e.g. flash
memories). In line with the recent development of the Internet,
search engines are mainly used to search information through the
Internet.
[0004] Although no official criterion has been established to
classify search engines, they are commonly divided into
subject-based search engines and keyword-based search engines
according to the operation type.
[0005] The subject-based search engines provide a list of
categories corresponding to major subjects of information available
from the Internet (e.g. society, culture, art, sports, politics).
Since they provide a list of various pieces of information
corresponding to the subject of interest, the subject-based search
engines are also referred to as directory servers, subject-based
catalogs, or menu search engines.
[0006] The subject-based search engines are advantageous in that,
when a user cannot pick a specific subject word or keyword leading
to the desired information, he/she can easily access the relevant
information. However, this type of access to information requires a
number of stages, such as "large category middle category small
category desired information." If an erroneous path is followed in
the middle of the search, the user may deviate from the desired
information.
[0007] In contrast, the keyword-based search engines are
advantageous in that only a small number of keywords (search words)
are enough to find the desired information quickly. When keywords
are entered, a list of search results is provided. However, if the
search word is ambiguous or if the search word has different
meanings depending on the context, respective meanings are not
differentiated, but are intermingled in the search results (lists
and excerpts of documents, images, photos, audios, video, flashes,
etc.).
[0008] It will be assumed for example that, in order to search
English contents related to bushes, a user enters "bush" as the
search word. Then, a conventional keyword-based search engine will
provide a list of contents, in which the search results regarding
"Bush" (biographical name) are intermingled with those regarding a
"bush" (in the sense of a shrub). In this case, the user will have
some difficulty in finding the desired contents.
[0009] Furthermore, when the entered search word is ambiguous or
has different meanings depending on the context, and when some of
the different meanings of the search word occupy the majority of
top-ranking contents of the search results, the user must review
almost all search results until he/she reaches the desired contents
that are ranked very low. This is unfavorable in terms of both time
and efficiency.
[0010] In short, conventional keyword-based search engines have a
problem in that, when the search word is ambiguous or has different
meanings depending on the context, the different meanings are
intermingled with one another in the search results. As a result,
the user must spend considerable time and energy until he/she finds
the desired information.
SUMMARY OF THE INVENTION
[0011] Accordingly, the present invention has been made to solve
the above-mentioned problems occurring in the prior art, and the
present invention provides a system and a method for automatically
classifying search results so that, when a search word entered to a
search engine adopting a conventional keyword-based search method
is ambiguous or has different meanings depending on the context,
the problem of intermingling of different meanings in the search
results is avoided.
[0012] The present invention also provides a system and a method
for automatically classifying search results so that, when a search
word entered to a search engine adopting a conventional
keyword-based search method is ambiguous or has different meanings
depending on the context, the problem of intermingling of different
meanings in the search results is avoided, thereby guaranteeing
that the user can find the desired information quickly and
efficiently.
[0013] In accordance with an aspect of the present invention, there
is provided a system for automatically classifying search results,
the system including a search engine server 10 for obtaining and
providing search results 50 with regard to a search word entered by
a user, receiving grouped search results 50, and providing the user
with the grouped search results 50; a related word database 20 for
storing related words classified into groups according to meanings
of the search word; and a group determination system 30 for
receiving search results 50 from the search engine server 10,
comparing the search results 50 with the related words stored in
the related word database 20 to determine which group of the
related word database 20 the search results 50 belong to, providing
the search engine server 10 with grouped search results 50, and
storing search results 50 at a predetermined place when the search
results 50 are not grouped which means that either the search
results 50 are belong to no group of the related word database 20
or the search results 50 belong to all of the groups.
[0014] Preferably, the system further includes a category
determination system 40 for classifying the search results 50
stored at the predetermined place without being grouped by the
group determination system 30 according to domain names providing
the search results 50.
[0015] Preferably, the search engine server 10 includes a search
result query processor for querying the group determination system
30 regarding which group the search results 50 corresponding to the
search word entered by the user belong to.
[0016] Preferably, the group determination system 30 includes a
count processor for counting how many related words stored in a
plurality of groups constituting the related word database 20 are
contained in contents of the search results group by group; a group
allocation processor for determining which group of the related
word database 20 the search results belong to according to a number
counted by the count processor; and a non-group allocation
processor for identifying search results 50 not being grouped and
storing the identified search results 50 at a predetermined
place.
[0017] Preferably, the group determination system 30 includes a
count processor for counting how many related words stored in a
plurality of groups constituting the related word database 20 are
contained in contents of the search results 50 group by group; an
intelligent group decision processor for assigning weights to
respective related words stored in the related word database 20
according to a degree of correlation between the related words and
meanings of the search word; a group allocation processor for
combining the weights assigned to the related words by the
intelligent group decision processor with numbers counted by the
count processor to determine which group of the related word
database 20 the search results 50 belong to; and a non-group
allocation processor for identifying search results 50 not being
grouped and storing the identified search results 50 at a
predetermined place.
[0018] Preferably, the category determination system 40 includes a
domain group database for storing domain names classified
hierarchically into at least two groups; and a category-based
search result allocation processor for classifying the search
results 50 by determining which group of the domain group database
a host belongs to according to a domain name of the host, the host
having provided the search results.
[0019] Preferably, the system further includes a user preference
acceptance system for enabling the user to set the number of
columns, the grouped search results 50 being outputted in the
columns, and redisplaying the search results according to the user
setting or displaying next search results 50 according to the user
setting; and an initial column number update system for
statistically surveying user preferences regarding the number of
groups, the search results being classified into the groups, and
automatically updating an initial setting according to the user
preferences.
[0020] In accordance with another aspect of the present invention,
there is provided a method for automatically classifying search
results with reference to a related word database storing groups of
words related to a search word, the method including the steps of
(a) receiving a search word entered by a user (S10); (b) obtaining
search results with regard to the entered search word (S20); (c)
grouping the search results by a group determination system with
reference to the related word database (S30, S40); and (d) sorting
and providing the grouped search results in at least two groups
(S50).
[0021] Preferably, in step (c) (S30, S40), weights are assigned to
respective related words belonging to each group of the related
word database according to a degree of correlation between the
related words and meanings of the search word, each weight is
combined with a number of appearance of the related word in the
search results, and the search results are allocated to a group
having a high combined weight.
[0022] Preferably, the method further includes a step of (e)
repeating steps (c) and (d) when the user wants to divide some
groups of the grouped, outputted search results into subgroups and
terminating searches when the user does not want to divide some
groups of the grouped, outputted search results into subgroups.
[0023] Preferably, the method further includes a step of (f)
classifying the search results according to a group of a domain
group database, a domain name providing the search results
belonging to the group, by a category determination system when it
is considered impossible to group the search results by the group
determination system, and outputting the classified search
results.
[0024] The system and method for automatically classifying search
results according to the present invention are advantageous as
follows: search results are grouped with reference to the related
word database, which stores related words grouped according to the
meaning or usage of the search word, so that the results are
separately provided according to the meaning and field of interest.
As a result, the user can access desired search results more
quickly and accurately.
[0025] The system and method selectively employ the group
determination system, which compares search results with the
related word database and groups the search results accordingly, or
the category determination system, which groups the search results
based on the domain name of the host of the search results, so
that, even if search results are not grouped by the group
determination system, they can be grouped by the category
determination system. This substantially improves the convenience
and efficiency of searching.
BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The above and other objects, features, and advantages of the
present invention will be more apparent from the following detailed
description taken in conjunction with the accompanying drawings, in
which:
[0027] FIG. 1 shows the overall construction of a system for
automatically classifying search results according to the present
invention;
[0028] FIG. 2 is a flowchart showing a method for automatically
classifying search results according to the present invention;
[0029] FIG. 3 shows a process for grouping search results by a
system for automatically classifying search results according to
the present invention; and
[0030] FIG. 4 shows an exemplary screenshot when a system for
automatically classifying search results according to the present
invention has been applied.
DETAILED DESCRIPTION OF THE EXEMPLARY EMBODIMENTS
[0031] Hereinafter, an exemplary embodiment of the present
invention will be described with reference to the accompanying
drawings. In the following description and drawings, the same
reference numerals are used to designate the same or similar
components, and so repetition of the description on the same or
similar components will be omitted.
[0032] FIG. 1 shows the overall construction of a system for
automatically classifying search results according to the present
invention.
[0033] Referring to FIG. 1, the system for automatically
classifying search results according to the present invention
includes a search engine server 10 for receiving a search word
entered by the user, providing the user with obtained search
results 50, receiving an input of grouped search results 50, and
providing the user with the grouped search results 50; a related
word database 20 for storing groups of words related to respective
meanings of search words; and a group determination system 30 for
receiving search results 50 from the search engine server 10,
comparing the search results 50 with the related words stored in
the related word database 20 to determine which group of the
related word database 20 the search results 50 belong to, providing
the search engine server 10 with the grouped search results 50, and
storing the search results 50 at a specific place when the search
results 50 do not belong to any group of the related word database
20.
[0034] The search engine server 10 according to the present
invention obtains and provides search results 50 with regard to
keyword searches using the Internet, personal computers, networking
computers, or other available online/offline search devices. In
addition, the search engine server 10 receives an input of grouped
search results 50 and displays them or outputs them as audible
signals (e.g. voices, sounds).
[0035] The search engine server 10 according to the present
invention refers to a search engine server adapted to output search
results using any type of search device, including a search engine
server adapted to output results with regard to keyword searches of
Internet users, a search engine server adapted to output results
with regard to desktop keyword searches (desktop, local,
neighboring, wideband network, etc.) of computer users, and a
universal search engine server adapted to output results with
regard to keyword searches using mobile search devices (e.g. flash
memories).
[0036] The search engine server 10 according to the present
invention includes a search result query processor for querying the
group determination system 30 about search results 50. The search
result query processor is adapted to query the group determination
system 30 about search results 50, in which different meanings of
search words are intermingled.
[0037] The related word database 20 according to the present
invention stores groups of related words, which have been
classified according to the meaning of search words based on
consideration of the correlation between the frequency of
appearance of words in web pages or offline corpus and the meaning
of search words.
[0038] The related word database 20 according to the present
invention stores groups of words related to respective meanings of
search words, which are ambiguous or have different meanings
depending on the context. The groups of the related word database
20, which have been classified according to the meaning of search
words, can be further divided into subgroups according to the
extent to which the meanings of search words are correlated.
[0039] Particularly, a hierarchical clustering method is used to
divide the related word database 20 into subgroups. According to
this method, if the related word database has two groups, each
group is further divided into subgroups. This subdivision may be
based on a distance calculation method, which combines related
words having close meanings into one. However, this degrades the
speed.
[0040] Therefore, a number of related word databases 20 having
different numbers of groups with regard to a single initial search
word are preferably established so that the user can select a
related word database 20 having the desired number of groups. This
is favorable in terms of speed.
[0041] The system for automatically classifying search results
according to the present invention employs a related word database
20, which is provided by the system provider, to classify search
results 50. The detailed process or method for dividing the related
word database 20 into a number of groups lies out of the scope of
the present invention, and descriptions thereof will be omitted
herein.
[0042] The group determination system 30 according to the present
invention compares the contents of search results 50 resulting from
an entered search word with related words to determine which group
of related words the search results 50 belong to. As used herein,
the contents of search results 50 refer to a set of words within
the search results resulting from a keyword search on the web.
[0043] The group determination system 30 according to the present
invention includes a count processor for counting the number of
related words, which are both stored in the groups constituting the
related word database 20 and included in the contents of search
results 50 resulting from an entered search word, a group
allocation processor for determining which group of the related
word database the search results 50 belong to based on the number
counted by the count processor, and a non-group allocation
processor for recognizing search results, which are not grouped,
and storing the search results 50 at a predetermined place.
[0044] Preferably, the group determination system 30 according to
the present invention further includes an intelligent group
decision processor for assigning weights to respective related
words stored in the related word database 20 according to the
degree of correlation between their meaning and that of search
words. In this case, the weights assigned to respective related
words by the intelligent group decision processor are combined with
the number counted by the count processor to determine which group
the research results 50 belong to. This process efficiently groups
the search results 50.
[0045] The group determination system 30 according to the present
invention can group most search results. However, if it is
considered meaningless or impossible to group some search results
by the group determination system 30, a category determination
system 40 may be used to classify the search results.
[0046] The category determination system 40 according to the
present invention determines the category of search results 50
based on the domain name of the search results, when the group
determination system 30 determines that it is meaningless or
impossible to group the search results by the non-group allocation
processor.
[0047] The category determination system 40 according to the
present invention groups the contents of search results 50
independently of the group determination system 30. The category
determination system 40 includes a domain group database storing at
least two groups of domain names of Internet hosts, which have been
classified hierarchically, and a category-based search result
allocation processor for classifying search results by determining
which group of the domain group database the host, which has
provided the search results 50, belongs to based on the domain name
of the host.
[0048] The system for automatically classifying search results
according to the present invention may include a user preference
acceptance system for allowing the user to set the number of
columns, in which grouped search results are outputted, and
reflecting the setting to redisplay the search results or
reflecting the setting to display the next search results, and an
initial column number update system for statistically surveying
user preferences regarding the number of groups, into which search
results are classified, and automatically updating the initial
setting based on the user preferences.
[0049] When the user designates the desired number of groups and
starts a search, the user preference acceptance system according to
the present invention refers to a related word database 20, which
has the same number of groups as the user setting, and outputs the
search results 50 based on the same number of groups.
[0050] The initial column number update system according to the
present invention statistically surveys user preferences regarding
the number of groups, and automatically updates the initial setting
of the number of groups into which search results 50 are
classified. Therefore, the user of the system for automatically
classifying search results according to the present invention does
not have to enter the desired group number for every search, since
the system refers to the cookie, for example, and automatically
classifies the search results 50 into groups, the number of which
is favored by the user. This is the same case as the user of Google
Search who can determine the desired number of results to be
displayed per page (e.g. 10, 30, or 100 results per page) in the
"Preferences" menu.
[0051] It can be said that the system for automatically classifying
search results according to the present invention secondarily
groups search results 50 obtained by search results 10 in various
manners. Therefore, the system can be operated independently of the
search mode of the search engines 10. Those skilled in the art can
also easily understand that the system can be interlinked with and
operated together with a search engine 10 when the search engine 10
composes a search word index table, i.e. when the search engine 10
conventionally composes a search table regarding search words
before users enter search words and start the search process.
[0052] FIG. 2 is a flowchart showing a method for automatically
classifying search results according to the present invention.
[0053] Referring to FIG. 2, the method for automatically
classifying search results according to the present invention
refers to the related word database table, which stores groups of
words related to search words, and classifies search results
accordingly. The method includes a first step (S10) of receiving a
search word entered by the user, a second step (S20) of obtaining
search results with regard to the entered search word, a third step
(S30 and S40) of referring to the related word database and
grouping the search results by the group determination system, and
a fourth step (S50) of sorting the grouped search results in at
least two columns and providing them.
[0054] More particularly, when the user enters a search word
corresponding to desired information (S10), search results are
obtained with regard to the entered search word without classifying
them according to the meaning or usage of the search word (S20).
After the search results are obtained by the search engine server,
the search engine server provides the group determination system
with the obtained search results. The group determination system
then refers to the related word database, which stores groups of
related words according to the meaning of search words (S30), and
groups the search results. The search engine server is provided
with the grouped search results, which are outputted as video
signals or audio signals (e.g. voices, sounds) (S50).
[0055] The process for referring to the related word database to
group the search results and the process for outputting the grouped
search results as video signals or audio signals (e.g. voices,
sounds) will now be described in more detail with reference to
FIGS. 3 and 4.
[0056] FIG. 3 shows a process for grouping search results by a
system for automatically classifying search results according to
the present invention.
[0057] Referring to FIG. 3, the process for grouping search results
by a system for automatically classifying search results according
to the present invention proceeds as follows: it is primarily
determined to group the search results by the group determination
system (S41) with reference to the related word database, which
stores groups of related words (S42). The grouped search results
are then outputted (S43). If the group determination system cannot
group the search results, the category determination system (S45)
secondarily groups the search results based on the domain name of
the host of the search results and outputs them.
[0058] If the user wants to divide some groups of the grouped
search results into subgroups (S44), it is determined again if the
group determination system (S41) can subdivide the groups. When the
group determination system can subdivide the groups, the subdivided
search results are outputted (S43). When the group determination
system cannot subdivide the groups, the category determination
system (S45) divides some groups of the search results into
subgroups and outputs them (S46).
[0059] The system for automatically classifying search results
according to the present invention primarily uses the group
determination system (S41) to compare the related word database,
which stores groups of related words, with the contents of search
results and determine if grouping is possible (S42).
[0060] For example, when the user enters "bush" as the search word
and starts a search, words related to the "bush" are obtained as
follows:
[0061] George, Mr, tree, rose, administration, Clinton, trees,
green, grass, ground, Bill, wild, low, campaign, leaves, p., thick,
bird, congress, road, thorn, meeting, beating, covered, USA, rock,
visit, camp, beat, birds, garden, shepherd, growing, announced,
summit, Gorbachev, Iraq, talks, hill, June, republican, yards,
flowers, cattle, branches, burning, forest, Reagan, dense, edge,
presidential, Moses, fruit, plant, dry, Nov., July, decision,
address.
[0062] The related word database is divided into a number of groups
according to the meaning of the search word. Assuming that the
related word database is divided into two groups with regard to the
ambiguous search word "bush," the first group contains words
related to "Bush" as a biographical name, and the second group
contains words related to the "bush" in the sense of a shrub. More
particularly, the related word database is grouped as follows:
[0063] First group: Reagan, summit, bush, Moses, address, George,
Bill, meeting, Mr, visit, Iraq, USA, campaign, June, talks,
announced, decision, July, Nov., p., congress, Gorbachev, Clinton,
presidential, administration, republican; and
[0064] Second group: tree, rose, trees, green, grass, ground, wild,
low, leaves, thick, bird, road, thorn, beating, covered, rock,
camp, beat, birds, garden, shepherd, growing, hill, yards, flowers,
cattle, branches, burning, forest, dense, edge, fruit, plant,
dry.
[0065] The group determination system according to the present
invention determines which of the first and second groups of the
related word database has more words related to the contents of
search results.
[0066] If it is determined that only one of the groups of the
related word database has words related to the contents of search
results, the group determination system considers that the contents
of search results belong to the corresponding group. For example,
if the contents of search results have no related words belonging
to the first group, but only those belonging to the second group,
the group determination system considers that the contents of
search results belong to the second group.
[0067] If the contents of search results have related words
simultaneously belonging to at least two groups of the related word
database, the group determination system groups the search results
based on consideration of the number of appearance of related words
belonging to respective groups in the contents of search results
together with the priority of the related words.
[0068] More particularly, the group determination system according
to the present invention counts the number of appearance of related
words, which belong to the first group, in the contents of search
results, and that of related words belonging to the second group by
using the count processor. The counted numbers are compared to
determine the group having more related words appearing in the
contents of search results. The contents of search results are
considered belonging to the determined group.
[0069] In addition, the group determination system according to the
present invention can assign weights to respective related words
belonging to each group for calculation and determination.
Particularly, the intelligent group decision processor of the group
determination system assigns weights to respective related words
belonging to each group of the related word database according to
the degree of correlation between their meaning and that of search
words for decision.
[0070] Assuming for example that words which are related to the
above-mentioned search word "bush" and which belong to the first
group of the related word database are arranged in a
multi-dimensional space according to the degree of correlation with
the biographical name "Bush", closely related words are arranged at
the center of gravity, while those with little correlation are far
from the center.
[0071] Particularly, if a related word "Reagan" is located near the
center of gravity of the meaning of related words of the first
group while a related words "republican" is far from the center,
the related word "Reagan" is given a weight because it is more
likely to belong to the first group.
[0072] In addition, if the related word database has groups
classified according to the meaning or usage of search words and if
the groups lie adjacent to one another, some related words may be
located near boundaries far from the center of gravity of the
meaning of respective groups. Such related words make little
contributions to grouping, and thus are given very low weights.
[0073] Preferably, the group determination system considers both
the number of appearance of related words belonging to each group
in the contents of search results and the weights assigned to them
based on the meaning when making a decision. In other words, the
weight of related words is combined with the number of appearance
in the contents of search results, and the contents of search
results are considered belonging to the group having the highest
total weight. Such consideration of both the number of appearance
and the weight of related words guarantees that the contents of
search results are grouped in a more precise manner.
[0074] The group determination system according to the present
invention determines which group the search results belong by using
the count processor and the intelligent group decision processor.
The group determination system groups the search results by using
the group allocation processor or the non-group allocation
processor, and provides the search engine server with the grouped
search results.
[0075] When the contents of search results belonging to a specific
group of the related word database, the group allocation processor
according to the present invention allocates the search results to
the corresponding group. When the contents of search results is not
grouped, the non-group allocation processor stores the search
results at a predetermined place. The search results stored at the
predetermined place by the non-group allocation processor are
grouped by the category determination system (described later).
Alternatively, the search results that have not been grouped may be
outputted as a single group according to user selection.
[0076] When the group determination system cannot clearly determine
the group to which the search results belong because the search
results are at the boundary of both groups, the search results are
displayed to both groups. The order of displaying the search
results are different between both groups according to the priority
decided by the group determination system.
[0077] The search results grouped by the group determination system
according to the present invention are displayed and outputted
(S43) in at least two columns by an output device (e.g. monitor).
In the case of the above-mentioned example, search results related
to the biographical name "Bush" may be displayed in the left
column, and search results related to the "bush" in the sense of a
shrub may be displayed in the right column. When the search device
and search server are adapted to provide search results audibly,
not visually, the search results belonging to respective groups are
provided as separate audio signals.
[0078] Although it has been assumed in the above exemplary
description that the related word database is divided into two
groups with regard to the search word "bush" entered by the user,
the user can arbitrarily set the number of groups of the related
word database. In this case, respective groups of the related word
database are divided into subgroups so that search results are
divided into the number of groups selected by the user.
[0079] When the user wants to divide some groups of the grouped
search results into subgroups (S44), the system for automatically
classifying search results according to the present invention
determines if the group determination system (S41) can again
subdivide the groups.
[0080] The related word database has a number of groups classified
according to the meaning of search words, and respective groups are
adapted to be divided into subgroups according to the degree of
correlation of related words in terms of their meaning. Therefore,
when the user wants to divide some groups of the grouped search
results into subgroups, it is determined if the related words
belonging to the corresponding groups of the related word database
can be grouped by the group determination system based on the
related word database, which has again been divided into
subgroups.
[0081] When some groups of the grouped search results can be
divided into subgroups by the group determination system, the
groups are divided into subgroups, and the corresponding search
results are outputted. When it is determined meaningless or
impossible to divide the groups into subgroups by the group
determination system, the category determination system (S45)
secondarily divides some groups of the search results into
subgroups.
[0082] When it has been determined meaningless or impossible to
group the search results by the group determination system, and
when the system for automatically classifying search results
according to the present invention has been notified (S42) of the
meaningless of grouping by the non-group allocation processor, the
category determination system (S45) may secondarily group the
search results.
[0083] The category determination system according to the present
invention includes a domain group database storing at least two
groups of domain names, which have been classified hierarchically.
For example, the domain group database has a first group of domain
names, such as ".com" and ".biz", and a second group of domain
names, such as ".edu" and ".org".
[0084] The category determination system according to the present
invention may refer to a database, which stores categorized domain
names, to classify the contents of search results by using the
category groups. For example, the category determination system
refers to a categorized database, which stores
"http://www.nytimes.com" in the news site category,
"http://www.nature.com/nature" in the journal category, etc., to
classify the search results.
[0085] The category determination system according to the present
invention can group the search results and separately output them
(S46) by the category-based search result allocation processor for
classifying the search results by determining which group of the
domain group database the host, which has provided the search
results, belongs to based on the domain name of the host.
[0086] FIG. 4 shows an exemplary screenshot when a system for
automatically classifying search results according to the present
invention has been applied.
[0087] Referring to FIG. 4, the system for automatically
classifying search results according to the present invention
outputs search results, which have been grouped by the group
determination system and the category determination system, in at
least two columns 54 and 56.
[0088] The search results grouped by the system for automatically
classifying search results according to the present invention are
displayed by an output device (e.g. monitor) in at least two
columns. For example, the first search results 54 related to the
biographical name "Bush" in the above-mentioned example are
displayed in the left column, and the second search results 56
related to the "bush" in the sense of a shrub are displayed in the
right column. When the search device and search server are adapted
to provide search results audibly, not visually, the search results
belonging to respective groups are provided as separate audio
signals.
[0089] Those skilled in the art can easily understand that,
although FIG. 4 shows search results displayed in at least two
columns, the search results can be displayed in any manner as long
as the search results can be recognized group by group. For
example, the search results may be displayed in rows.
Alternatively, the search results may be displayed in respective
sections of the interior of a circle (i.e. in a pie type) or any
other closed loop.
[0090] In addition, although it has been assumed in the description
with reference to FIG. 4 that the search results are divided into
two groups, the search results may be divided into at least three
groups and outputted when the related word database has at least
three groups or when the search results classified by the group
determination system are again classified into subgroups.
[0091] As mentioned above, the system for automatically classifying
search results according to the present invention is advantageous
in that search results are grouped with regard to a search word,
which is ambiguous or which has different meanings depending on the
context, and are outputted accordingly so that the user can not
only conduct a search easily, but also efficiently find the desired
information from the search results.
[0092] To be more specific, when one of the leading search engines
conducts a search with regard to the above-mentioned search word
"bush", search results related to the "bush" in the sense of a
shrub occupy no more than two of the upper 200 results, and the
remaining 198 search results are related to the biographical name
"Bush". This means that, if the user wants to find search results
related to the "bush" in the sense of a shrub, he/she must waste
considerable time and energy to find just two results from 200
results.
[0093] In contrast, the system for automatically classifying search
results according to the present invention groups search results
according to the meaning or usage of the search word and outputs
the search results in two, three, or at least four columns so that
the user can easily find the group to which the desired search
results belong. This substantially reduces the time and energy
necessary for searching.
[0094] Such an intelligent and efficient search engine is also
favorable to search engine business providers. If the desired
search results are not ranked high, users will have difficulty in
finding them and get disappointed. In contrast, if search results
are grouped and displayed separately so that users can easily find
the desired search results, the competitiveness of the search
engine business provider will be substantially improved.
[0095] Furthermore, if a search word regarding a product, which is
advertised on the web, is ambiguous or has different meanings
depending on the context, the search results provided by
conventional search engines with regard to the product tend to be
ranked low among the entire search results. This means that the
search results are less likely to be viewed by users. If the system
for automatically classifying search results according to the
present invention is employed in this regard, search results are
grouped according to the meaning or usage of the search word and
are outputted group by group so that the users can easily access
the search results regarding the advertiser's product.
[0096] Although an exemplary embodiment of the present invention
has been described for illustrative purposes, those skilled in the
art will appreciate that various modifications, additions and
substitutions are possible, without departing from the scope and
spirit of the invention as disclosed in the accompanying
claims.
* * * * *
References